Abstract
The phenomena associated with the fracture of glass are more logically expressed by a distribution function giving the probability of failure when the tensile stress at a point reaches a given value than by the concept of an intrinsic strength which is a characteristic of the material. In practice it is found that this distribution function has appreciable values only on the surface of glass. Functional equations are developed relating the results to be expected in tension tests and in modulus of rupture tests with one or two loading edges on cylindrical rods. Experiments with acid‐polished rods of Pyrex‐brand chemical glass require a two‐term distribution function dn= (2.9 × 10−9S2+ 7.15 × 10−21.S8) dS to express the results found by a bending test with one loading edge (units, kilograms and millimeters), whereas results given by Bailey for lime‐glass rods not polished are expressed by the function dn= 7.3 × 10–32 (units, pounds and inches) for tests with two loading edges.

This publication has 3 references indexed in Scilit: